2015 Informs Annual Meeting

SC06

INFORMS Philadelphia – 2015

3 - The Impact of Facebook on Offline Sales: Evidence from the U.S. Automobile Industry Yen-yao Wang, Michigan State University, N270A Business College Complex, East Lansing MI 48824, United States of America,wangyen@broad.msu.edu, Anjana Susarla, Vallabh Sambamurthy This study examines the dynamic interactions between firm-generated content, user-generated content, and offline car sales in the U.S. automobile industry. We collected the official Facebook pages of 31 car companies in the U.S., and supplemented the data from these firms’ traditional media efforts, and offline car sales from 2009 to 2014. A panel vector autoregressive (PVAR) model is conducted to examine our research framework. Implications for researchers and managers are discussed. 4 - An Investigation of Factors that Influence the Success of Social Commerce Platforms Suning Zhu, szz0012@auburn.edu, Jiahe Song, Pei Xu This paper introduces a descriptive framework for understanding factors that influence the success of social commerce platforms. It identifies three main classes of influencing factors (self-expression, communication, and style) and characterizes the individual factors in each class.The resultant framework can be used by researchers for reference of theories and hypothesis generation, and by practitioners for benchmarking social commerce practices. SC06 06-Room 306, Marriott Computational Methods in Options Pricing and Portfolio Selection Sponsor: Financial Services Sponsored Session Chair: Liming Feng, Associate Professor, University of Illinois at Urbana-Champaign, 104 S Mathews Ave, Urbana, IL, 61801, United States of America, fenglm@illinois.edu Co-Chair: Xuewei Yang, Associate Professor, Nanjing University, School of Management and Engineering, #22, Hankou Road, Gulou District, Xuewei Yang, Associate Professor, Nanjing University, School of Management and Engineering, #22, Hankou Road, Gulou District, Nanjing, 210093, China, xwyang@nju.edu.cn, Xindan Li, Avanidhar Subrahmanyam Turbo warrants are options that can be called back when underlying prices reach a threshold. We find that investors treat turbo warrants like lotteries in that they prefer those with low prices, high volatilities, and high skewness, and prefer trading them when underlying prices are near callback thresholds. As a result, turbo warrants are overpriced: during 2012, investors lost 1.82 billion HKD (US$235 million) by trading turbo warrants written on the Hang Seng Index. 2 - Asymptotic Expansions of Discretely Monitored Barrier Options under Stochastic Volatility Models Chao Shi, Assistant Professor, Shanghai University of Finance and Economics, 100 Wudong Road, Yangpu District, Shanghai, 200433, China, shichao@connect.ust.hk, Ning Cai, Chenxu Li This paper proposes an algorithm for pricing discretely monitored barrier options under stochastic volatility models. It turns out that the celebrated Hilbert transform recursion algorithm proposed by Feng and Linetsky (2008) becomes the leading term and building block in our expansion formula under stochastic volatility models. Our expansions are automatic and fast. Numerical results show that our algorithm is efficient and robust. 3 - Distributions with Analytic Characteristic Functions in Financial Modeling Runqi Hu, PhD Student, University of Illinois at Urbana- Champaign, 104 S Mathews Ave, Urbana, IL, 61801, United States of America, runqihu2@illinois.edu, Liming Feng In this talk, we consider a class of distributions with characteristic functions that are analytic in a horizontal strip in the complex plane. Such distributions can be inverted from their characteristic functions very efficiently using simple rules with exponentially decaying approximation errors. The results can be applied in accurate valuation of options in models with jumps and stochastic volatility. Numerical examples illustrate the effectiveness of the schemes. Nanjing, 210093, China, xwyang@nju.edu.cn 1 - Investor Behavior and Turbo Warrant Pricing

4 - Robust Portfolio Selection with Fixed Transaction Cost Yufei Yang, PhD Candidate, Singapore University of Technology and Design, Pillar of Engineering Systems and Design, 8 Somapah Road, Singapore, 48732, Singapore, yufei_yang@mymail.sutd.edu.sg, Selin Damla Ahipasaoglu, Jingnan Chen We study a robust mean-variance portfolio selection problem under fixed transaction cost. We provide a novel analysis on the portfolio composition and a closed-form formula is derived to unify various types of portfolios. We uncover the impact of the uncertainty level and fixed transaction cost to the position change of each asset.

SC07 07-Room 307, Marriott Big Risks, Big Data Cluster: Risk Management Invited Session

Chair: Paul Glasserman, Columbia Business School, 3022 Broadway, Uris Hall, New York, United States of America, pg20@columbia.edu 1 - Large-dimensional Factor Modeling Based on High-frequency Observations Markus Pelger, Assistant Professor, Management Science & Engineering, Stanford University, Huang Engineering Center, 475 Via Ortega, Stanford, CA, 94305, United States of America, markus.pelger@gmail.com I provide a statistical theory and empirically estimate an unknown factor structure based on financial high-frequency data for a large cross-section. I develop an estimator for the number of factors and consistent and asymptotically mixed- normal estimators of the loadings and factors for a large number of cross-sectional and high-frequency observations. In an extensive empirical study of the U.S. equity market I identify four continuous and one jump factor that explain most of the variation. 2 - Price Contagion through Balance Sheet Linkages Agostino Capponi, Columbia, Mudd 313, New York, NY, 10027, United States of America, ac3827@columbia.edu We study price linkages between assets held by financial institutions that maintain fixed capital structures over time. Our analysis suggests that regulatory policies aimed at stabilizing the system by imposing capital constraints on banks may have unintended consequences: banks’ deleveraging activities may amplify asset return shocks and lead to large fluctuations in realized returns. We show that these effects can be mitigated by encouraging banks to hold liquid, rather than illiquid, assets. 3 - Incorporating GICS and High-Frequency Data into Portfolio Allocation and Risk Estimation Jianqing Fan, Princeton, Dept of Operations Res & Fin Eng, Princeton University, Princeton, NJ, 08544, United States of America, jqfan@princeton.edu, Alexander Furger, Dacheng Xiu We document a striking block-diagonal pattern in the factor residual covariances of the S&P 500 constituents, after sorting the assets by their assigned GICS codes. We propose combining a location-based thresholding approach based on sector inclusion with the Fama-French and SDPR sector ETF’s. We investigate the performance of our estimators in a portfolio allocation study. We provide justification for the empirical results by jointly analyzing the in-fill and diverging dimension asymptotics. 4 - Estimating the Correlation Matrix of Credit Default Swaps for Market Risk Management Richard Neuberg, Columbia University, 1255 Amsterdam Avenue, Dept of Statistics, 10th Floor, New York, NY, 10027, United States of America, rn2325@columbia.edu, Paul Glasserman We propose a portfolio perspective to better understand the properties of correlation matrix estimators and loss functions for market risk management. We find the commonly used latent factor model to systematically misestimate the risk of certain portfolios. The normal likelihood appears more appropriate than Frobenius’ and Stein’s loss. We derive specific loss functions. We assess a range of estimators using CDS data. We also study implied CDS correlations using distance-to-default processes.

94